Aerodynamic Stability and Performance of Next-Generation ...

AIAA Aerodynamic Decelerator Systems (ADS) Conference 

25-28 March 2013, Daytona Beach, Florida 

AIAA 2013-1356 

Aerodynamic Stability and Performance of Next-Generation 

Parachutes for Mars Descent 

Keir C. Gonyea 1 

Georgia Institute of Technology, Atlanta, GA, 30332 

Christopher L. Tanner 2 and Ian G. Clark 3 

Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA, 91109 

Laura K. Kushner 4 and Edward T. Schairer 5 

NASA Ames Research Center, Moffett Field, CA 94035 

Downloaded by GEORGIA INST OF TECHNOLOGY on April 1, 2013 | http://arc.aiaa.org | DOI: 10.2514/6.2013-1356 

Robert D. Braun 6 

Georgia Institute of Technology, Atlanta, GA, 30332 

The Low Density Supersonic Decelerator Project is developing a next-generation 

supersonic parachute for use on future Mars missions. In order to determine the new 

parachute configuration, a wind tunnel test was conducted at the National Full-scale 

Aerodynamics Complex 80- by 120-foot Wind Tunnel at the NASA Ames Research Center. 

The goal of the wind tunnel test was to quantitatively determine the aerodynamic stability 

and performance of various canopy configurations in order to help select the design to be 

flown on the Supersonic Flight Dynamics tests. Parachute configurations included the diskgap-band, 

ringsail, and ringsail-variant designs referred to as a disksail and starsail. During 

the wind tunnel test, digital cameras captured synchronized image streams of the parachute 

from three directions. Stereo photogrammetric processing was performed on the image data 

to track the position of the canopy vent throughout each run. The position data were 

processed to determine the geometric angular history of the parachute, which was then used 

to calculate the total angle of attack and its derivatives at each instant in time. Static and 

dynamic moment coefficients were extracted from these data using a parameter estimation 

method involving the one-dimensional equation of motion for the rotation of a parachute in 

a wind tunnel. The coefficients were calculated over all of the available canopy states to 

reconstruct moment coefficient curves as a function of total angle of attack. From the 

stability curves, useful metrics such as the peak moment, trim total angle of attack, and pitch 

stiffness at the trim angle could be determined. The moment curves, aided by the stability 

metrics, were used to assess the parachute’s stability in the context of its drag load and 

geometric porosity. While there was generally an inverse relationship between the drag load 

and the stability of the canopy, the data showed that it was possible to obtain similar stability 

properties as the disk-gap-band with slightly higher drag loads by appropriately tailoring 

the geometric porosity distribution. 

________________________ 

1 

Graduate Research Assistant, School of Aerospace Engineering, keir@gatech.edu, AIAA student member 

2 

Lead Test Engineer, christopher.l.tanner@jpl.nasa.gov, AIAA member 

3 

LDSD Principal Investigator, ian.g.clark@jpl.nasa.gov, AIAA member 

4 

Research Engineer, ACI/Experimental Aero-Physics Branch, laura.k.kushner@nasa.gov 

5 

Aerospace Engineer, Experimental Aero-Physics Branch, edward.t.schairer@nasa.gov 

6 

David and Andrew Lewis Professor of Space Technology, School of Aerospace Engineering, 

robert.braun@aerospace.gatech.edu, AIAA Fellow 

1 

American Institute of Aeronautics and Astronautics 

This material is declared a work of the U.S. Government and is not subject to copyright protection in the United States.


Nomenclature 

C m 

C m0 

static moment coefficient 

local intercept of static moment curve 

C mα 

local slope of static moment curve 

C m α 

dynamic moment coefficient 

C T 

tangential force coefficient 

D 0 

parachute reference diameter 

D p 

parachute projected diameter 

g 

acceleration due to gravity 

k 44 

apparent inertia coefficient 

m 

mass of the parachute canopy 

M aero 

moment due to canopy aerodynamics 

Q w 

local dynamic pressure at the canopy 

R cm 

distance from the ball joint to the canopy center of mass 

R cp 

distance from the ball joint to the canopy center of pressure 

R v 

distance from the ball joint to the canopy vent 

S 0 

parachute reference area 

V c 

wind velocity at the canopy 

V w 

wind velocity at the canopy corrected for canopy rotation 

V t 

velocity of the canopy tangent to its arc of motion 

x, y, z wind tunnel frame coordinates (x streamwise, y lateral, z vertical) 

α 

angle of attack 

α G 

total geometric angle 

α T 

total angle of attack 

β 

sideslip angle 

Δα, Δβ dynamic contribution to the angle of attack and sideslip angle, respectively 

γ 

geometric angle between V c and V t 

φ 

clock angle (angle from vertical of wind tunnel axis projected onto yz-plane, positive clockwise) 

ρ ∞ 

freestream air density 

θ, ψ geometric pitch and yaw angles, respectively 

Ω 

magnitude of the angular velocity of the canopy 

Subscripts 

v 

θ 

ψ 

trim 

location of the canopy vent 

motion in the pitch plane 

motion in the yaw plane 

trim total angle of attack 

Superscripts 

’ parachute body axes 

Acronyms 

LDSD 

DGB 

DS 

NFAC 

PIA 

RMS 

RS 

SS 

TDT 

Low Density Supersonic Decelerator 

disk-gap-band 

disksail 

National Full-scale Aerodynamics Complex 

Parachute Industry Association 

root mean square 

ringsail 

starsail 

Transonic Dynamics Tunnel 

2 




T 

I. Introduction 

he Low Density Supersonic Decelerator (LDSD) project is developing a next-generation supersonic parachute 

to be considered for use on future Mars missions. The resulting canopy design is expected to update or replace 

the disk-gap-band (DGB) parachute that has flown on all previous U.S. missions to the surface of Mars. Many 

canopy variations were considered including ringsail and DGB parachutes as well as new designs referred to as 

disksail and starsail parachutes in order to understand the effects of distributing porosity throughout a canopy. 1 

LDSD quantified the stability characteristics of each canopy design through wind tunnel testing of sub-scale 

canopies (approximately 35% scale) with representative gore and ring structures. Static and dynamic aerodynamic 

coefficients (C m and C m α 

, respectively) were estimated for each canopy as a function of total angle of attack (α T ). 

The aerodynamic coefficient curves were used to obtain stability metrics such as the peak moment, trim total angle 

of attack, and slope of the static aerodynamic curve at the trim total angle of attack for each canopy. These metrics 

help quantify the stability of each parachute so that they may be compared relative to one another. 

Stability is an important factor in overall parachute performance. Chaotic motions of a parachute have the 

potential to disrupt guidance algorithms used to control the entry vehicle during descent and risk causing system 

instability. However, experimental determination of parachute aerodynamics is difficult because they are highly 

flexibly structures, have complex flow interactions, and exhibit apparent mass effects. A test was conducted in the 

NASA Langley Transonic Dynamics Tunnel (TDT) that was able to characterize some these effects by holding a 

textile parachute at the vent and rotating the parachute-payload system through a range of angles of attack. 2 While 

this test was technically more accurate than previous experiments using rigid parachute models, the error resulting 

from artificially holding the parachute at a constant angle of attack was not quantified. Moreover, this method of 

testing is not feasible in larger facilities such as NFAC due to the cost of constructing the necessary moving fixtures. 

A second portion of the TDT test involved using a free flying parachute to determine drag performance. 2 A few 

years after the completion of the test, Schoenenberger et al. used video data from a downstream camera to extract 

the parachute stability coefficients. 3 By tracking the location of the canopy vent in each video frame and 

transforming those data into a two-dimensional position in space, the total angle of attack and its first and second 

derivatives could be computed. These values were subsequently used in a parameter estimation methodology to 

calculate the static and dynamic aerodynamic coefficients at a given total angle of attack. The aerodynamics 

calculated for the parachute correlated well with the static test results for the same canopy. 2 This parameter 

estimation methodology outlined in Ref. 2 is well suited to large-scale parachute testing and is the primary method 

being used to resolve the parachute stability characteristics. 

Since the conversion of video data into parachute aerodynamics was not a primary objective of the TDT 

experiment, several approximations had to be made in order to compensate for the lack of some pieces of data. In 

particular, the use of a single downstream video camera caused ambiguity in the parachute location and the rapid 

motion of the canopy relative to the video frame rate induced error in the calculation of the angular rates and 

accelerations. The LDSD wind tunnel test attempted to increase the knowledge of the parachute position by utilizing 

stereo photogrammetry and calculation of the angular derivatives was improved with data acquisition occurring at 

60 Hz. 

II. Test Setup 

A. Canopy Description 

LDSD tested a total of 4 different canopy types and a total of 13 different configurations. 1 The test articles had a 

nominal diameter (D 0 ) of 11.8 m (38.8 ft) and a suspension line length of 1.7D 0 . The majority of the canopies were 

constructed from PIA-C-44378 “F-111” nylon broadcloth, which has a fabric permeability of less than 5 ft 3 /min/ft 2 

per its specification. For the canopies constructed from F-111 nylon, the total porosity is assumed to be equal to the 

geometric porosity since the contribution from fabric porosity is assumed to be negligible. 

Two main parameters were varied in the test articles: the magnitude of the geometric porosity and the 

distribution of the geometric porosity. Generally, higher drag canopies tend to be less stable; thus any improvement 

in stability from an increase in geometric porosity is expected to be coupled with a reduction in drag. However, it is 

hypothesized that intelligent modifications to the geometric porosity distribution can balance the increase in stability 

with a minimal reduction in drag. To accomplish this goal, sail panels were removed from rings located at various 

distances from the canopy skirt to determine if it was advantageous to preference the geometric porosity distribution 

near or away from the skirt. In addition, different circumferential porosity distributions were investigated by 

removing either a full ring or every other panel. The canopy designs that were tested are discussed below. Note that 

higher number rings are located further away from the canopy apex (closer to the canopy skirt). 

3 




1) Disk-gap-band: DGB canopies are constructed by separating a flat circular disk and a cylindrical band of 

fabric by an open gap to aid in stability. The DGB canopy serves as the reference by which all of the nextgeneration 

parachutes are assessed. Two configurations were tested: 

a. DGB-1: a flight spare of the parachute used for the Mars Phoenix Scout lander mission, 

constructed using MIL-C-7020 Type I nylon, which has a permeability of approximately 100 

ft 3 /min/ft 2 . For this canopy, the contribution from fabric porosity is non-negligible and the total 

porosity was calculated to be between 12-18%. 

b. DGB-2: a replica of the Phoenix DGB constructed using F-111 nylon. This test article is shown in 

Fig. 1a. 

2) Ringsail: ringsail parachutes are modifications of ringslot parachutes that add fullness to the fabric panels 

and allow for more airflow through the canopy. Five configurations were tested: 

a. RS-0: a subscale version of a Ringsail parachute tested by JPL in 2005. 4 A picture of this test 

article is shown in Fig. 1b. 

b. RS-1: the RS-0 canopy with two-thirds of ring 19 removed. 

c. RS-2: the RS-0 canopy with 27% of rings 17, 18 and 19 removed. 

d. RS-3: the RS-0 canopy with all of ring 19 removed. 

e. RS-4: the RS-0 canopy with all of rings 18 and 19 removed. 

3) Disksail: the disksail canopy is a modification of the Ringsail canopy that replaces the first ten rings around 

the canopy vent with a flat circular disk. The goal of this configuration was to decrease geometric porosity in 

the crown of the parachute to increase drag and allow that porosity to be redistributed to other portions of the 

canopy. Five configurations were tested: 

a. DS-1: the disksail as described above and as shown in Fig 1c. 

b. DS-2: the DS-1 canopy with half of ring 11 removed. 

c. DS-3: the DS-1 canopy with all of ring 11 removed. 

d. DS-4: the DS-1 canopy with all of ring 11 and half of ring 17 removed. 

e. DS-5: the DS-1 canopy with all of ring 11 and half of rings 17 and 18 removed. 

4) Starsail: the starsail canopy is a modification of the Ringsail where multiple gores are replaced with a solid 

material creating a star pattern. The goal of this configuration is to change how the geometric porosity is 

distributed throughout the canopy to retain drag and obtain some desirable stability characteristics. Portions 

of rings 17-20 were removed to obtain a geometric porosity approximately equal to the DGB. One starsail 

configuration was tested and is shown in Fig. 1d. 

(a) disk-gap-band (b) ringsail (c) disksail (d) starsail 

Figure 1. Primary canopy configurations used in NFAC testing. 

Each canopy was equipped with fourteen retro-reflective targets on both sides of 

the canopy that appeared in high contrast against the test article and allowed for the 

canopy to be more easily tracked by the photogrammetry system described in Section 

II.C. Fiducial target material was carefully selected to maximize light return across a 

relatively broad range of incidence angles. Targets were located in three concentric 

rings around the vent with coded target patterns on the outer-most ring to resolve 

parachute roll about its axis of symmetry. The target pattern is shown in Fig. 2. 

B. Test Conditions 

The wind tunnel testing was performed at the National Full-scale Aerodynamics 

Complex (NFAC) 80- by 120-foot (80x120) Wind Tunnel at the NASA Ames 

Research Center. Parachutes were fixed to a strut at the center of the test section via a 

Figure 2. Retro-reflective 

target pattern on each 

test article. 

4 



load arm and ball joint. Mounted to the front of the strut was an aeroshell simulator, which was intended to 

approximate the wake generated by the forebody that will be present during future flight tests. This aeroshell 

simulator was fixed to the strut and was not allowed to move with the parachute. A diagram of the test setup can be 

seen in Fig. 3. 

The canopies were tested at nominal freestream wind velocities of both approximately 15 and 25 kts. Pressure 

probes measured the dynamic pressure during the test and were located both upstream of the strut to measure the 

freestream conditions and downstream of the canopy skirt to measure blockage effects. 

C. Photogrammetry System 


1. Photogrammetry Setup 

The purpose of the photogrammetry system was accurately measure the position of the test articles in threedimensional 

space to be used in estimating their static and dynamic stability characteristics. The photogrammetry 

hardware consisted of three high-resolution (2352x1728 pixels) synchronized cameras, two downstream of the 

parachute on the floor of the test section diffuser and one upstream of the parachute mounted on the strut just below 

tunnel centerline. The locations of the cameras and the choice of lenses were determined using virtual-imaging 

software to predict the camera views and ensure that the canopies would be visible over the expected range of 

positions. 5 The two downstream cameras were placed symmetrically near the corners of the test section to provide 

stereo imaging of the outer surface of the canopy. They were located sufficiently far downstream to be able to view 

the retro-reflective targets on the canopy at up to 20° total angle of attack in any direction. The upstream camera was 

mounted just below the riser attachment and provided a full view of the inside surface of the canopy. The cameras 

acquired images at 60 Hz – more than ten times the oscillation frequency of the parachute, thereby eliminating any 

aliasing of the canopy motion. High-intensity lamps were placed next to each camera to maximize the light output of 

the retro-reflective targets on the canopy and minimize the uncertainty in the position tracking. The photogrammetry 

configuration relative to the overall test set-up can be seen in Fig 3. A synchronized view from each of the 

photogrammetry cameras is shown in Fig. 4. 

Figure 3. Planview of the wind tunnel test section. 

2. Photogrammetry Calibration 

The biggest challenge in making photogrammetry measurements on such a large scale was calibrating the cameras. 

Therefore, two independent calibration methods were used, which provided verification for each other. The first and 

simplest method was the Direct Linear Transformation, which required first placing and focusing the cameras and 

then imaging at least six targets in the region of interest whose spatial coordinates were known. 6 The second method 

required first measuring the “internal orientation” of each camera (focal length, lens distortion corrections, and 

location of the optical axis in the image plane) before the cameras were mounted. This was accomplished by 

acquiring images with each camera of a planar array of known targets. These targets were applied in a rectangular 

grid to one sidewall of the test section. Then, after the cameras were mounted in their final positions and pointed, the 

spatial positions and point angles of the cameras (“external orientation”) were computed from images of a set of 

targets in the fields of view whose spatial coordinates were known. 

5 



(a) View from west camera (b) View from east camera (c) View from strut camera 


Figure 4. Synchronized images from the three photogrammetry camera views. Stereo photogrammetric 

measurements were computed using the east and west views. 

Calibration targets were placed on a crane positioned in the region of interest, the strut fairing, and the test 

section sidewalls. The space coordinates of the calibration targets were precisely determined by imaging them from 

many directions using a commercial photogrammetry system. Both the Direct Linear Transform and 

internal/external orientation methods resulted in coefficients for each camera, which, together with image-plane 

coordinates of targets that appear in the images of at least two cameras, allowed computation of the space 

coordinates of the targets. Unlike the single-camera measurements used in Ref. 3 and previous photogrammetry 

measurements of parachutes in the 80x120, 7 the stereo imaging method used for this test allowed for accurate threedimensional 

tracking of the vent without assuming a constant distance from the canopy to the point of rotation. 

3. Photogrammetry Validation 

The uncertainty in the photogrammetry system was determined by comparing the camera measurements of 

verification targets against their known coordinates. Measurements were made with the targets supported on a lift at 

three different heights and three different lateral locations at the streamwise position of the canopies. The relative 

error of the photogrammetry measurements was determined by first translating and rotating the measured 

coordinates of the targets to minimize the root mean square (RMS) difference with the true coordinates. The 

resulting minimum RMS error was less than half of an inch. The uncertainty in the absolute position of the targets 

was estimated by dangling a tape measure and plumb bob from the rig to the floor of the test section and then 

measuring to known reference points. Based on these measurements, the uncertainty in absolute position was less 

than one inch. These uncertainty estimates are consistent with the expected uncertainty due to a one-pixel error in 

locating targets in the images. The spatial position of the vent was calculated using both the Direct Linear 

Transformation and the internal/external calibration methods, resulting in similar coordinates. The internal/external 

calibration method was ultimately selected to generate all of the data herein. 

III. Data Analyses 

A. Canopy Vent Coordinates to Geometric Angles 

Once the position history of the canopy was determined, 

the coordinates of the vent were converted into geometric 

angles, which are more convenient for describing the 

rotational motion of the parachute. Geometric angles are 

defined here as angles that are dependent only on the 

parachute’s position with respect to the wind tunnel and do 

not take into account the parachute’s motion with respect to 

the wind. A diagram showing the wind tunnel and parachute 

reference frames as well as the relevant geometric angles is 

shown in Fig. 5. The wind tunnel frame is denoted as {x, y, z} 

and the parachute frame is denoted as {x’, y’, z’} with the 

origin located at the ball joint, The parachute angular 

velocity is defined as Ω. The parachute and wind tunnel 

frames are related by a series of Euler rotations, first by the 

pitch angle (θ) about the y-axis, followed by the yaw angle 

(ψ) about the z’-axis. The full rotation matrix can be seen in 

Eq. (1). 

Figure 5. Wind tunnel and canopy coordinate 

systems. 

6 



! 

# 

# 

# 

" 

x 

y 

z 

$ ! 

& # 

& = # 

& 

% "# 

cosθ cosψ −cosθ sinψ sinθ $ ! 

& x' $ 

# & 

sinψ cosψ 0 &# 

y' & 

−sinθ cosψ sinθ sinψ cosθ # 

%& 

" z' & 

% 

(1) 

The length of the parachute from the ball joint to the vent is defined as R v . Knowing R v and the {x v , y v , z v } 

coordinates of the canopy vent, the pitch and yaw angles can be calculated via Eqs. (3) and (4). 


R v 

= x v 2 + y v 2 + z v 

2 

" 

θ = sin −1 z 

− v 

% 

$ ' (3) 

# R v 

cosψ & 

" 

ψ = sin −1 

$ 

# 

y v 

R v 

(2) 

% 

' (4) 

& 

Two other geometric angles that are convenient to define are the total geometric angle (α G ) and clock angle (φ). 

The total geometric angle is the total angular distance between the parachute x’-axis and the wind tunnel x-axis. 

Note that the total geometric angle is not the same as the total angle of attack, which will be defined later. The clock 

angle describes the parachute position in the yz-plane when looking upstream. It is defined to be φ = 0 when y v = 0 

and z v > 0 and φ = π/2 when z v = 0 and y v > 0. The total geometric angle and the clock angle can be calculated via 

Eqs. (5) and (6). 

" 

α G 

= cos −1 

$ 

# 

x v 

R v 

% 

' (5) 

& 

" 

φ = tan −1 sinθ cosψ % 

$ ' (6) 

# sinψ & 

1. Calculating the Total Angle of Attack and its Derivatives 

The total angle of attack can be expressed in terms of the traditional angle of attack and sideslip angle as in Eq. 

(7). Note that the total angle of attack is always positive due to its physical definition. 

α T 

= cos −1 

[ cosα cosβ] (7) 

If the canopy is stationary, then the angle of attack is equal to the pitch angle, the sideslip angle is equal to the 

yaw angle, and the total angle of attack is equal to the total geometric angle. However, if the parachute is moving, 

then the rotational motion alters the local wind velocity at the canopy and introduces dynamic contributions 

(Δα, Δβ) to the geometric pitch and yaw angles, as in Eqs. (8). 

α =θ + Δα (8.1) 

β = ψ + Δβ (8.2) 

Calculating the aerodynamic coefficients requires knowledge of the first and second derivatives of the total angle 

of attack with respect to time, which can be calculated using finite differencing. However, since α T is always 

positive, its value can change very rapidly around zero and potentially create non-smooth derivatives. An analytic 

method of calculating the first and second derivatives of the total angle of attack was developed that only requires 

finite differencing of the Euler angles θ and ψ. These angles have both positive and negative magnitudes and vary 

smoothly in time, making them well suited for differentiation via finite differencing. The first and second derivatives 

of the total angle of attack are given in Eqs. (9) and (10). Additional details regarding the calculation Δα, Δβ, and 

their respective derivatives are given in Appendices A and B. 

7 



α T 

= α sinα cosβ + β cosα sin β 

sinα T 

(9) 

α T 

= α sinα cosβ + β cosα sin β + ( α 2 + β 2 − α 2 T 

)cosα T 

− 2 α β sinα sin β 

(10) 

sinα T 


B. Local Wind Velocity at the Canopy 

The total wind velocity at the canopy is the vector sum of the freestream wind velocity (V c ) and the wind 

velocity tangent to the canopy’s arc of motion (V t ). Note that the wind velocity tangent to the canopy’s arc of 

motion is equal and opposite to the tangential velocity of the canopy, thus it is subtracted from the V c as in Eq. 

(11.1). The total wind velocity (V w ) is the magnitude (L 2 - norm) of the total wind velocity vector (V w ) given in Eq. 

(11.2). 

V w 

= 

V w 

= V c 

− V t 

(11.1) 

( V c 

− x cp ) 2 + y 2 2 

cp 

+ z cp 

(11.2) 

The velocity of the canopy tangent to its arc of motion can be expressed in terms of the Euler angles (see Fig. 5) 

as in Eq. (12). The canopy velocity is taken at the center of pressure (R cp ), which is where the aerodynamic forces 

are assumed to act. 

! 

# 

V t 

= # 

# 

" 

# 

x cp 

y cp 

z cp 

$ ! 

& 

− 

$ 

# 

θ sinθ cosψ − ψ cosθ sinψ 

& 

& = R cp 

# ψ cosψ & 

& # 

% 

& 

" 

−θ 

& 

# cosθ cosψ + ψ sinθ sinψ 

% 

& 

C. Calculating the Aerodynamic Coefficients 

The aerodynamic moments on the parachute are represented as a static moment, dependent on the parachute’s 

total angle of attack, and a dynamic moment, dependent on the instantaneous rate of change of the total angle of 

attack. The static moment curve is locally linearized at each total angle of attack into the pitch stiffness ( C mα 

) and 

the moment at 0° total angle of attack ( C m0 

), as in Eq. (13.1). The resulting expression for the total aerodynamic 

moment is given in Eq. (13.2) where Q w is the dynamic pressure accounting for canopy rotation, S 0 is the parachute 

reference area, and D 0 is the parachute reference diameter. 

(12) 

C m 

= C mα 

α T 

+ C m0 

(13.1) 

! D 

M aero 

= Q w 

S 0 

D 0 

C 0 

$ 

# m α 

α T 

+ C mα 

α T 

+ C m0 & (13.2) 

" 2V w 

% 

The angular behavior with respect to the wind (described in Section III.A) can be used to determine the canopy 

stability coefficients using parameter estimation. 3 Given that the parachute is an axisymmetric body, the entire 

attitude history can be decomposed into motion in two directions - in the same direction as the total angle of attack 

and in the direction orthogonal to the total angle of attack. It is assumed in this analysis that the time-averaged 

aerodynamic coefficients in the direction orthogonal to the total angle of attack are zero. For motion in the same 

direction as the total angle of attack, the rotational equation of motion of the parachute in a wind tunnel can be 

expressed as in Eq. (14), which accounts for forcing due to aerodynamic moments and gravity. I yy is the moment of 

inertia of both the canopy and the apparent mass, m is the mass of the canopy only, and g is the gravitational 

acceleration. Equation (14) can be rearranged to explicitly solve for the aerodynamic moment coefficients as seen in 

Eq. (15). 

8 



! 

"# 

C m α 

C mα 

C m0 

I yy 

a T 

= Q w 

S 0 

D 

! 

0 

C m α 

C mα 

C 

"# 

m0 

$ 

%& = ' 1 * 

) , I yy 

a T 

( Q w 

S 0 

D 0 + 

! 

# 

$ # 

%& # 

# 

" 

α T 

D 0 

2V w 

α T 

1 

$ 

& 

& 

& 

& 

% 

( [ ] − mgR cp [ cosφ cosα T ]) 

+ mgR cp [ cosφ cosα T ] (14) 

! 

# 

# 

# 

# 

" 

α T 

D 0 

2V w 

α T 

1 

T 

$ !! 

& ## 

& ## 

& ## 

& ## 

% "# 

" 

α T 

D 0 

2V w 

α T 

1 

$ ! 

&# 

&# 

&# 

&# 

%" 

α T 

D 0 

2V w 

α T 

1 

−1 

T 

$ $ 

& & 

& & 

& & 

& & 

% %& 

(15) 


Equation (15) was simultaneously solved across a small range (or bin) of total angles of attack in order to obtain 

a set of coefficients that are representative of the parachute behavior within that α T range. This bin was then 

incrementally stepped across the full range of α T data in order to obtain a relatively smooth curve relating the 

moment coefficients to the total angle of attack. The resultant coefficients are assumed to correspond to the average 

total angle of attack within each bin. The use of a larger bin size will result in a smoother curve, but it will tend to 

bias the resulting coefficients towards those angles of attack that occurred the most. The increment at which the bin 

is moved controls the density of points along the curve. Also, the upper and lower bounds of the moment curves are 

limited by the angles that were traversed by the parachute during testing and the bin size selected. 

D. Discussion of the Apparent Mass 

Parachute aerodynamics are often hard to analyze because of complex interactions with the surrounding 

flowfield. For example, when a parachute is moving in a fluid, any external force that accelerates the parachute must 

also accelerate the fluid in and around the canopy. This fluid acceleration can be thought of as an additional mass of 

the system and is often referred to as the apparent mass. The effect of the apparent mass is very difficult to isolate 

since it is dependent on the fluid density, canopy size, canopy porosity, flow compressibility, and flow velocity. The 

apparent mass is often mathematically described as a 6x6 tensor with values based in both potential flow theory and 

empirical data. 8 

Ibrahim 9 performed a series of experiments to quantify the apparent inertia of rotating hemispherical, flat circular, 

guide surface, and ribbon canopies. 9 For each of the canopies, he determined a non-dimensional coefficient of the 

apparent moment of inertia for rotation around the canopy centroid as well as rotation around the canopy confluence 

point. The apparent inertia coefficient was non-dimensionalized with respect to a sphere of air of diameter equal to 

the projected diameter of the canopy as seen in Eq. (16). Apparent inertias ranged from approximately 31% of a full 

sphere of air for a hemispherical canopy to 9% for a ribbon canopy. Uncertainty in these inertias was not 

documented. 

I yy 

k 44 

= 

(16) 

1 2 

6 

π D p3 

ρ ∞ 

R cm 

Given the relatively small weight of the canopies in this test and the high-density air at sea-level, the apparent 

inertia about the ball joint dominates the I yy term in the present analysis. Since the gravity term in Eq. (15) is much 

smaller than the aerodynamics term, the moment coefficients are approximately proportional to the apparent inertia. 

As a result, the apparent mass acts as a scaling factor on the calculated moment coefficients. This is a particularly 

important point since, as stated above, the correct apparent inertia value is very difficult to determine and the error in 

the calculated moment coefficients will be magnified by the error in the apparent inertia. Therefore, the results for 

the moment coefficients in Section IV are presented given the current best estimate of the apparent inertia. 

IV. Results 

Photogrammetric data was acquired for each canopy, although only a representative set of data are presented 

herein. For discussion purposes, Figs. 6-8 are presented for the RS-1 canopy at the 25 kt test condition. However, 

similar trends were also seen for the other canopies and conditions. 

9 




A. Two-Dimensional Canopy Motion 

Figure 6 shows a two-dimensional trace of the canopy motion 

in the wind tunnel yz-plane (plane perpendicular to the wind tunnel 

centerline). The dots along the curve represent data at 3 Hz and 

illustrate that the 60 Hz data rate provided a sufficiently dense 

sampling of the canopy motion. It can be seen that the parachute 

stays approximately within a circle of radius twenty feet, centered 

near the tunnel centerline. In addition, the parachute covers the 

entire interior of the circle fairly uniformly, showing that the 

canopy never develops a circular coning motion near its trim angle 

of attack. The parachute’s time-averaged position in the y-direction 

is negligible and shows that there was no tendency for it to stay on 

either side of the test section. However, the average position in the 

z-direction is noticeably below zero, which can be attributed to 

gravity acting on the canopy. 

Figure 6. Trace of the RS-1 canopy vent. 

B. Dynamic Versus Static Angle Contribution 

The result of the total angle of attack calculation (described in Section III.A) is displayed in Fig. 7. Figure 7a 

shows that the wind-relative angles are significantly greater than the geometric angles due to rotation of the canopy. 

The mean and 95 th percentile α G and α T are shown in Fig. 7b and indicate that the wind-relative angles can be 

greater than twice the geometric angles. Figure 7b also shows that the distributions of the angles change 

considerably. This is particularly important since the stability curves, which should be calculated based on total 

angle of attack, would look significantly different if based off of the total geometric angle. 

(a) Angular vent trace of the canopy 

motion over time 

(b) Histogram of angles traversed by the parachute 

Figure 7. Comparisons of the angular motion of the parachute when using geometric angles and windrelative 

angles for the RS-1 canopy. 

In addition, the use of wind-relative angles leads to a non-intuitive relationship between the total geometric angle 

and the total angle of attack. Figure 8a shows the tangential velocity of the canopy versus the total geometric angle 

at each point in the parachute trajectory. The tangential velocity is generally high at low total geometric angles and 

low at high angles. Thus, the parachute momentarily stops rotating when it reaches the maximum total geometric 

angle and rotates the fastest as it sweeps through the center, similar to simple harmonic motion. This means that the 

parachute reaches its largest total angle of attack just after passing through the center of the test section (α G near 

zero). It then reaches the lowest total angle of attack just after attaining the maximum total geometric angle, while 

10 



eginning to return to the center of the test section. In other words, the maximum and minimum total geometric 

angle and the total angle of attack are approximately 180° out of phase from each other. This behavior can be seen in 

Fig. 8b. 


(a) Tangential velocity vs. total geometric 

angle 

C. Raw Data Reduction and Processing 

The stability coefficients were determined using a bin width of 0.5° and a bin step of 0.25°. The bin width was 

chosen because there were generally over 25 points contained within this bin size, which was assumed to be a 

sufficiently large sample size to generate representative coefficients. The bin step was chosen to provide an adequate 

number of data points from which to reconstruct the continuous C m curve. A plot of the resultant C m data for the 

RS-1 canopy is shown in Fig. 9 (both the blue circle and purple x symbols). These data were curve fit using an 8 th 

order polynomial that was forced to go through a C m of zero at 0° total angle of attack (which is typical of 

axisymmetric bodies). The data appeared to exhibit an unusually high C m at low total angles of attack, thus some 

data were excluded from the fit to obtain a reasonable C m curve, which are seen as the purple x symbols in Fig. 9. 

These curve fits will be used for the relative comparison of different canopies, although their absolute magnitudes 

may not be accurate due to the uncertainty in the apparent mass value used in the analysis. This topic is discussed 

further in Section IV.E. 

The trim total angle of attack is the angle where the parachute does not experience an aerodynamic moment (C m 

is equal to 0). A low trim angle of attack is desirable since it will be less likely to introduce a destabilizing moment 

on the payload and because more of the drag force will be 

oriented along the centerline of the payload. For canopy 

RS-1, there are two trim angles – 0° and 23° total angle 

of attack. The positive moment curve slope at 0° is 

indicative of an unstable trim point, where a small 

perturbation will force the canopy away from the trim 

total angle of attack. Conversely, the negative moment 

curve slope at 23° indicates a stable trim point, where any 

deviation of the parachute from this point will drive it 

back to the trim total angle of attack. The magnitude of 

C mα,trim 

determines the magnitude of the restorative force, 

or how stable the parachute is at the trim total angle of 

attack. While a low trim total angle of attack is always 

considered beneficial, it is not clear what is the best value 

for C mα,trim 

. If moment curve slope is too low, then the 

restorative force is relatively weak and the parachute may 

traverse large angles during descent. However, if the 

moment curve slope is too large, then the parachute could 

(b) Total angle of attack vs. total geometric 

angle 

Figure 8. Comparisons between the tangential velocity and the total angle of attack profiles 

to the total geometric angle for the RS-1 canopy. 

Figure 9. Static moment coefficients and curve fit as a 

function of the total angle of attack for the RS-1 

canopy. X symbols were excluded when performing 

the curve fit. 

11 




potentially introduce a large, violent moment on the payload 

if it were suddenly displaced from the trim total angle of 

attack due to a gust of wind or other perturbation. Another 

important feature of the curve is the peak C m value. Higher 

peak values could also potentially cause violent motion and 

could cause destabilizing system dynamics. Therefore, a 

lower overall C m curve is considered to be beneficial. 

Figure 10 shows a plot of the pitch damping curve for the 

RS-1 canopy. In this case, the pitch damping coefficient at 

the trim total angle of attack is less than zero; therefore, the 

canopy is dynamically stable at this point. However, the 

curves of different canopies vary widely and there is no 

overall trend regarding their dynamic stability at the trim 

total angle of attack. To obtain a smooth curve it was 

necessary to increase the bin size to 1.5°. However, the 

coefficients values still scatter towards lower total angles of 

attack. As a result, it is difficult to determine the shape of 

the pitch damping curve, which makes comparisons between 

canopies difficult. 

D. Comparison to Heritage Wind Tunnel Results 

Prior to the Mars Exploration Rover missions, wind 

tunnel tests of various DGB parachutes were performed in 

the TDT to determine their drag performance and static 

stability behavior. 2 Moment values for each canopy were 

measured by constraining the parachute in a fixture that 

was rotated through a range of angles of attack. The data 

from this test have served as the basis of the parachute 

aerodynamics models for all subsequent U.S. Mars 

missions. In addition, the success of the DGB parachutes 

used in these missions demonstrates that these data are 

representative of Mars flight conditions and are the closest 

aerodynamics set to true parachute motion currently 

available. Therefore, it is useful to compare the results of 

the present NFAC test to the TDT test to ensure that the 

aerodynamics predicted by each test are not in conflict. 

As part of the TDT test campaign, a sub-scale version of 

the Mars Viking DGB was flown that had a nominal 

diameter of approximately 5.2 ft and was constructed from 

Figure 10. Dynamic moment coefficients as a 

function of total angle of attack for the RS-1 

canopy. 

Figure 11. Comparison of C m curves as a function of 

total angle of attack for wind tunnel tests performed 

in the NFAC and the TDT. 

MIL-C-7020 Type III nylon. This test was run at sea-level density and a dynamic pressure of 16 psf. This canopy is 

very similar to the Mars Phoenix Scout canopy (DGB-1) flown in the present NFAC test since the Phoenix DGB gap 

and band heights were based on the Viking configuration and the fabric permeability of Type I and Type III 

MIL-C-7020 nylon are similar. The two NFAC DGB-1 tests were conducted at dynamic pressures of 0.8 and 2.5 psf. 

Figure 11 shows the resulting C m curves from each of the tests. 

Comparison between the TDT and NFAC tests is difficult because the runs were performed at very different 

dynamic pressures. However, it can be seen that the trim total angle of attack decreases with increasing dynamic 

pressure, which was similarly observed from the TDT testing. 2 Additionally, the peak C m and the general shape of 

the C m curves appear to change with the dynamic pressure. 

E. Apparent Mass Effects 

Equation 16 shows that the apparent inertia model scales with the parachute nominal diameter to the fifth power 

(given that the distance R cm is a function of the nominal diameter). Assuming that the error in the apparent inertia is 

a constant percentage its nominal value, error in the apparent mass model would be significantly greater for large 

diameter parachutes than for small parachutes. Uncertainty related to the apparent inertia of the parachute canopies 

tested at NFAC, as stated in Section III.D, may be one potential cause of the differing C m curves shown in Fig. 11. 

12 




One apparent inertia coefficient value of 0.05 was used 

to analyze all of the canopies, despite that they had 

different geometric porosity and were operated at slightly 

different dynamic pressures. The data set in Ref. 9 does not 

provide enough data to intelligently select an apparent 

inertia coefficient that depends on geometric porosity and 

dynamic pressure. The lowest apparent inertia coefficient 

cited in Ref. 9 was 0.087, which corresponded to a ringslot 

canopy with a geometric porosity of approximately 27%. 

However, this apparent inertia coefficient resulted in a C m 

curve that differed significantly from the existing DGB data, 

as shown in Fig. 12. As such, a value of 0.05 was used, 

which provided a slightly better correlation with the 

existing DGB data. 

F. Comparison Between Canopy Aerodynamics 

Figure 12. C m curves calculated using varying 

The stability metrics for each canopy are tabulated in apparent inertia coefficients for the DGB-1 canopy 

Table 1 along with their averaged tangential force 

compared to heritage data. 

coefficient (C T ) and approximate geometric porosity. 

Desirable canopies have low trim total angles of attack and high averaged tangential force coefficients. 

Canopy 

Number 

Canopy Description 

Table 1. Summary of canopy stability and drag results. 

1. Disk-gap-band Comparison 

Figure 13 shows the static stability curves for the DGB- 

1 and DGB-2 canopies at the same dynamic pressure. 

While both canopies have the same geometric porosity, the 

DGB-1 has a higher total porosity (15-18%) than DGB-2 

due to higher fabric permeability. Figure 13 indicates that, 

for the two DGB parachutes tested, higher fabric 

permeability effectively decreases the peak C m value, the 

trim α T , and the tangential force. The TDT test was 

conducted with DGB’s having two different material 

permeabilities as well and trim α T was similarly observed 

to decrease. 2 Since DGB parachutes have displayed 

acceptable stability behavior during prior U.S. Mars 

missions, the overall performance of each parachute can be 

determined in relation to the performance of the DGB (for 

example, equivalent stability with enhanced drag). 

Geometric 

Porosity (%) 

Trim α T 

(deg) 

(1/deg) 

Averaged 

C T 

DGB-1 DGB with high porosity fabric 13 8 -6 x10 -3 0.59 

DGB-2 DGB with low porosity fabric 13 15 -9 x10 -3 0.81 

RS-0 Ringsail design tested in 2005 10 23 -6 x10 -3 0.99 

RS-1 RS-0 without 2/3 ring 19 13 23 -8 x10 -3 0.90 

RS-2 RS-0 without 27% rings 17, 18, 19 15 24 -7 x10 -3 0.91 

RS-3 RS-0 without ring 19 16 21 -8 x10 -3 0.86 

RS-4 RS-0 without rings 18, 19 22 19 -11 x10 -3 0.77 

DS-0 Disksail as built 9 23 -9 x10 -3 1.03 

DS-1 DS-0 without 1/2 ring 11 11 19 -8 x10 -3 0.98 

DS-2 DS-0 without ring 11 13 13 -15 x10 -3 0.92 

DS-3 DS-0 without ring 11, 1/2 ring 17 16 12 -13 x10 -3 0.86 

DS-4 DS-0 without ring 11, 1/2 rings 17, 18 19 14 -10 x10 -3 0.82 

SS Starsail as built 13 23 -5 x10 -3 0.83 

Figure 13. Comparison of C m curves for the DGB 

canopies. 

13 




2. As-built Canopy Comparisons 

Figure 14 shows a comparison of the moment 

coefficient curves for the F-111 DGB and the as-built 

ringsail, disksail, and starsail canopies. The DS-0 and RS-0 

canopies have similar C m profiles and similar trim total 

angles of attack, but the disksail exhibits slightly better 

drag performance than the ringsail. In fact, Table 1 

indicates that other disksail canopies have a smaller trim α T 

and equivalent or greater tangential force than ringsail 

configurations with similar geometric porosities. 

Additionally, the C m curves for disksail canopies tend to 

have a steeper slope around the trim total angle of attack 

than for ringsails with equivalent geometric porosities. It is 

unclear why this trend occurs, but it was evident during 

testing that the disksail took a slightly more blunt shape 

than the ringsail, which was hypothesized to have occurred 

because of the presence of the flat disk in the crown. 

Figure 14. Comparison of the C m curves for the 

unmodified canopies. 

The starsail canopy has a similar trim total angle of attack to the RS-0 and DS-0 but has less tangential force. 

However, the starsail C m curve is different since is lower than the RS-0 and DS-0 C m curves and has a relatively 

shallow slope at the trim total angle of attack. In this sense, the starsail is more neutrally stable than the ringsail or 

disksail. However, given that the disksail and ringsail canopies had the same trim total angle of attack and much 

higher drag, the starsail was considered to be a less effective design. It should also be noted that the unconventional 

design of the starsail would have made it considerably more difficult to manufacture than the other two 

configurations. Therefore, the starsail experiment was not pursued further than the one configuration. 

3. Ringsail Comparisons 

The RS-1 and RS-2 canopies were designed to have similar geometric porosity, but with different geometric 

porosity distributions. The RS-1 canopy concentrated the geometric porosity all to ring 19, where it was hoped that a 

strong circumferential jet of air flowing out from the canopy would create uniform flow disruption and increase 

stability (similar to the design of a DGB). The RS-2 canopy distributed the porosity evenly between rings 17, 18, 

and 19, where it was hoped that the distributed porosity would induce different sized vortices and increase stability. 

However, manufacturing tolerances and a rushed fabrication schedule resulted in the RS-1 and RS-2 canopies 

having different geometric porosities. In general, the stability of the RS-0, RS-1, and RS-2 are similar, although the 

peak value of the C m curve is slightly different for the each canopy, as shown in Fig. 15a. In addition, the RS-1 and 

RS-2 canopies produced similar tangential force coefficients, which was approximately 10% lower than the RS-0 

canopy. Therefore, the change in the geometric porosity distribution around the shoulder region of the canopy had a 

relatively minimal effect. 

(a) Differing geometric porosity distribution 

(b) Differing geometric porosity magnitude 

Figure 15. Comparison of C m curves for the Ringsail canopy modifications. 

14 



The magnitude of the geometric porosity was intentionally modified in the RS-3 and RS-4 configurations, 

increasing the geometric porosity of the RS-0 canopy by approximately 60% and 120%, respectively. All of the 

porosity was created in the shoulder of the parachute to determine if a larger gap would improve the stability more 

than in the RS-1 and RS-2 configurations. Figure 15b shows that these changes in the geometric porosity did have a 

noticeable effect and decreased the trim total angle of attack by 9% and 17% for RS-3 and RS-4, respectively. 

However, the RS-3 and RS-4 canopies also exhibited a 13% and 22% decrease in the average tangential force 

coefficient relative to the RS-0 canopy. In addition, neither RS-3 nor RS-4 exhibited improved tangential force and 

stability behavior relative to DGB-2. 


4. Disksail Comparisons 

All of the alternate disksail configurations were obtained by successively removing sail panels from the DS-0 

canopy. As seen in Fig. 16a, the first two modifications (DS-1 and DS-2) have the smallest increase in total porosity, 

but cause the highest reductions in the trim total angle of attack relative to DS-0. Furthermore, configuration DS-2 

exhibits a similar trim total angle of attack to the DGB-2 but has a significantly higher tangential force coefficient 

and a slightly steeper C m curve at the trim α T . Further increases in the geometric porosity near the shoulder of the 

disksail in configurations DS-3 and DS-4 decrease the tangential force but do not significantly alter the trim 

behavior from the DS-2 configuration, as seen in Fig. 16b. From these data, it appears as if increasing porosity near 

the crown of the disksail (as in DS-1 and DS-2) causes the greatest decrease in the trim total angle of attack for the 

corresponding decrease in the tangential force. 

(a) Successive modifications to ring 11 (b) Successive modifications to rings 17 and 18 (on 

top of the ring 11 modifications) 

Figure 16. Comparison of C m curves for the Disksail canopy modifications. 

V. Conclusion 

Wind tunnel testing of various parachute configurations was performed to identify the relative drag and stability 

behavior of canopies with different geometric porosity magnitudes and distributions. Photogrammetric imaging of 

the canopies during testing was used to track the canopy vent and accurately determine its position in the test section 

to within one inch of uncertainty. Geometric and wind-relative angles were calculated from these photogrammetry 

data. Due to oscillatory motion of the canopy during testing, it was necessary to correct the aerodynamic angles to 

include dynamic as well as static components. This correction led to a non-intuitive total angle of attack profile. 

A parameter estimation methodology was used to extract static and dynamic moment coefficients as a function of 

the total angle of attack. This methodology was found to be especially sensitive to uncertainty in the apparent inertia 

model. Since it was not possible to measure the apparent inertia of the canopies, the apparent inertia was modeled 

based on historical work and data correlation. Moment coefficients were statistically estimated at every 0.25° α T and 

the data was curve fit using an 8 th -order polynomial. Some moment coefficients at low total angles of attack, where 

data was generally sparse, were selectively excluded to obtain a better fit. Stability metrics such as the trim angle of 

attack and slope at the trim angle were determined using these curve fits to aid in the comparison of the various 

canopy configurations. 

The behavior of the ringsail, disksail, and starsail canopies were compared against the DGB. The data showed 

that alteration of the geometric porosity in the shoulder region of the ringsail canopy did not yield tangential loads or 

15 



stability behavior that were more attractive than that of the DGB. Disksail configurations DS-2 and DS-3, however, 

exhibited significantly greater tangential force and equivalent stability behavior relative to the DGB. The starsail 

exhibited different behavior from the ringsail and disksail, but it did not appear to improve upon the DGB 

performance. Selection of a final canopy design is not presented due to lack of uncertainty analyses and the absence 

of data from a final wind tunnel test entry, which was not processed in time for this publication. 

Appendix 


A. Calculating the Total Angle of Attack 

For ease of explanation, the current discussion assumes that motion is restricted to the pitch plane, although the 

theory is applied similarly to motion in the yaw plane. Canopy rotation about the ball joint in the pitch plane is 

shown in Figs. 17a and 17b. Rotation of the canopy results in a velocity component that is tangent to the canopy’s 

circular arc of motion (V t ). Aerodynamic forces act through the center of pressure of the parachute, which is 

generally located near the skirt of the canopy (R cp ). Since the canopy forces are computed with respect to the total 

angle of attack and the velocity of the canopy, the velocity and total angle of attack are calculated at the center of 

pressure. The tangential velocity is given in Eq. 17. The velocity V c is the wind velocity at the canopy, which is 

assumed to act along the tunnel centerline and have a larger magnitude than the freestream wind velocity due to 

blockage effects. The resulting velocity triangles seen in Figs. 17c and 17d give rise to the actual wind velocity (V w ) 

at the center of pressure of the canopy and a dynamic angular component of the angle of attack. Note that the sign of 

the pitch rate ( θ ) depends on whether the canopy is rotating away from the tunnel centerline (positive) or towards 

the centerline (negative). Additionally, note that the direction of the tangential velocity in Figs. 17c and 17d is equal 

and opposite of that shown in Figs. 17a and 17b because the wind velocity with respect to the canopy is equal and 

opposite of the velocity of the canopy with respect to the wind. 

V t 

= R cp θ (17) 

(a) Canopy rotation for > 0 (b) Canopy rotation for < 0 

(c) Velocity triangle for > 0 (d) Velocity triangle for < 0 

Figure 17. Diagram of canopy rotation and the resulting wind velocity 

triangle for motion in the pitch plane. 

The angle γ is defined as the angle between V c and V t and is given in Eq. (18). In Eq. (18), “sgn” is the sign of the 

function and is equal to +1 when the pitch rate is positive and -1 when the pitch rate is negative. The actual wind 

velocity V w is then calculated via the Law of Cosines as in Eq. (19). 

γ = π 2 + sgn ( θ )θ (18) 

V w 2 = V c 2 +V t 2 − 2V c 

V t 

cosγ (19.1) 

16 



V w 

= 

V 2 2 

c 

+ R 

cp 

θ 2 − 2V c 

R " 

cp 

θ cos π 2 + sign ( θ % 

)θ 

# 

$ 

& 

' (19.2) 

With V w , V t , and γ known, the dynamic angle of attack correction Δα can be calculated via the Law of Sines as in 

Eq. (20). The angle of attack is then the sum of the geometric pitch angle and the dynamic correction, as in Eq. (21). 


# 

% 

Δα = sin −1 % 

% 

% 

$ 

sin Δα 

= sinγ θ 

(20.1) 

V tθ 

V wθ 

& 

sin π ( 

# 

2 + sgn ( θ &( 

)θ 

$ 

% 

' 

( 

(20.2) 

( 

( 

' 

α =θ + Δα (21) 

R cp θ 

V 2 2 

c 

+ R 

cp 

θ 2 − 2V c 

R # 

cp 

θ cos π 2 + sgn ( θ & 

)θ 

$ 

% 

' 

( 

Angle correction in the yaw plane is similar to the correction in the pitch plane. Therefore, the aerodynamic 

sideslip angle and its dynamic correction are given Eq. (22). 

# 

% 

Δβ = sin −1 % 

% 

% 

$ 

& 

R cp 

ψ 

V 2 c 

+ R 2 

cp 

ψ 2 − 2V c 

R cp 

ψ cos π sin π ( 

# 

# 

2 + sgn & 2 + sgn & 

ψ 

( 

( )ψ 

$ 

% 

' 

( 

(22.1) 

( 

( ψ )ψ 

$ 

% 

' 

( 

( 

' 

β = ψ + Δβ (22.2) 

The total angle of attack, accounting for both static and dynamic contributions, is given in Eq. (23). Note that the 

total angle of attack is always positive due to its physical definition. 

cosα T 

= cosα cosβ (23.1) 

α T 

= cos −1 

[ cosα cosβ] (23.2) 

B. Calculating Derivatives of the Total Angle of Attack 

The derivative of the total angle of attack can be calculated by taking the derivative of Eq. (23.1). 

α T 

= α sinα cosβ + β cosα sin β 

sinα T 

(24) 

The derivative of the angle of attack and the sideslip angle are equal to the sum of the derivatives of the static 

and dynamic components, as in Eq. (25). Derivatives of the dynamic contributions are given in Eq. (26). 

V 

Δ α = tθ 

V wθ 

−V tθ 

V wθ 

2 

V wθ 

α = θ + Δ α (25.1) 

β = ψ + Δ β (25.2) 

sinγ θ 

cosΔα + V t θ 

V wθ 

cosγ θ 

cosΔα γ θ (26.1) 

Δ β = 

V tψ 

V wψ 

−V tψ 

V wψ 

2 

V wψ 

sinγ 

ψ 

cosΔβ + V t ψ 

cosγ 

ψ 

V wψ 

cosΔβ γ ψ 

(26.2) 

17 



The derivatives of the tangential canopy velocity in the pitch plane ( γ θ 

) and the actual wind velocity in the pitch 

plane can be found by differentiating Eqs. (17), (18) and (19.1) and are calculated via Eqs. (27.1.1), (27.1.2), and 

(27.1.3) respectively. The derivatives of the tangential canopy velocity in the yaw plane ( γ ψ ) and the actual wind 

velocity in the yaw plane can be found in the same way and are calculated via Eqs. (27.2.1), (27.2.2), and (27.2.3) 

respectively. 

V tθ 

= R 

cp 

θ (27.1.1) 

γ θ 

= θ 

(27.1.2) 


V wθ 

= 

V wψ 

= 

V tθ ( V tθ 

−V c 

cosγ θ ) + γ θ 

V c 

V tθ 

sinγ θ 

(27.1.3) 

V wθ 

V tψ 

V tψ 

= R 

cp 

θ (27.2.1) 

γ ψ 

= ψ (27.2.2) 

( V tψ 

−V c 

cosγ 

ψ ) + γ 

ψ 

V c 

V tψ 

sinγ 

ψ 

(27.2.3) 

V wψ 

The second derivative of the total angle of attack can be calculated by twice differentiating Eq. (23.1). The 

remaining derivation proceeds in the same fashion as for the first derivative (given in Eqs. (24) through (27)). 

α T 

= α sinα cosβ + β cosα sin β + α 2 + β 2 2 

( − α T )cosα T 

− 2 α β sinα sin β 

(28) 

sinα T 

Δ α = 

Δ β = 

V 

wθ 

= 1 

α = θ + Δ α (29.1) 

β = ψ + Δ β (29.2) 

( )( V wθ ) − ( V tθ 

V wθ 

−V Vwθ 

 

tθ )( 2V Vwθ 

 

wθ ) 

# V V −V 2 

V 

% t θ wθ tθ wθ 

sinγ 

4 θ 

+ 

1 % 

V wθ 

% 

cosΔα % V 

2 tθ 

V wθ 

−V tθ 

V wθ 

γ 

2 θ 

cosγ θ 

+ V t θ 

% 

γ 2 θ 

sinγ θ 

+ γ θ 

cosγ θ 

$ V wθ 

V wθ 

# V V −V 2 

( V t ψ wψ t ψ wψ )( V wψ ) − ( 

V tψ 

V wψ 

−V tψ 

V wψ ) 2V wψ 

V 

% 

( wψ ) 

sinγ 

4 ψ 

+ 

1 % 

V wψ 

% 

cosΔβ % V tψ 

V wψ 

−V Vwψ 

 

tψ 

2 

γ 

2 

ψ 

cosγ ψ 

+ V t ψ 

% 

γ 2 ψ 

sinγ ψ 

+ γ ψ 

cosγ ψ 

$ % V wψ 

V wψ 

V wθ 

V 

wψ 

= 1 

V wψ 

" 

# 

Vtθ 2 + V tθ 

V tθ 

−V c 

cosγ θ 

" 

Vtψ 

 

# 

2 + 

V tψ 

& 

( 

( 

( 

(30.1) 

( ) + Δ α 2 ( 

sinΔα( 

' 

& 

( 

( 

( 

(30.2) 

( ) + Δ ( 

β 2 sinΔβ( 

'( 

V 

t θ 

= R 

cp 

θ (31.1.1) 

γ θ 

= θ (31.1.2) 

( ) + 2 γ θ 

V 

cVtθ sinγ θ 

+V c 

V tθ ( γ θ 

sinγ θ 

+ γ 2 θ 

cosγ θ ) −V 2 

wθ 

$ 

% (31.1.3) 

V 

t ψ 

= R cp 

ψ (31.2.1) 

γ ψ 

= ψ (31.2.2) 

( V tψ 

−V c 

cosγ ψ ) + 2 γ ψ 

V 

cVtψ sinγ ψ 

+V c 

V tψ 

γ ψ 

sinγ ψ 

+ γ 2 ψ 

cosγ ψ 

( ) −V 2$ 

wψ 

% (31.2.3) 

18 



C. Local Wind Velocity at the Canopy 

The parachute center of pressure can be expressed in the inertial frame via the transformation matrix in Eq. (1). 

The inertial coordinates of the center of pressure are found in Eq. (32). 

" 

$ 

R cp 

= R cp $ 

# $ 

cosθ cosψ % 

' 

sinψ ' 

−sinθ cosψ 

&' 

(32) 

The inertial angular velocity vector (Ω) of the canopy can be determined by rotating the Euler angle rates back to 

the inertial frame, as in Eq. (33). 


" 

$ 

Ω = $ 

$ 

# 

cosθ 0 sinθ 

0 1 0 

−sinθ 0 cosθ 

%" 

' $ 

' $ 

' $ 

&# 

0 

θ 

0 

% " 

' $ 

' + $ 

' $ 

& # 

cosθ 0 sinθ 

0 1 0 

−sinθ 0 cosθ 

! 

# 

Ω = # 

# 

" 

# 

ψ sinθ 

$ 

θ 

& 

& 

& ψ cosθ 

% 

& 

%" 

cosψ −sinψ 0 %" 

' $ 

' $ 

' $ sinψ cosψ 0 ' $ 

' 

&# 

$ 0 0 1 $ 

&' 

# 

0 

0 

ψ 

% 

' 

' 

' 

& 

(33.1) 

(33.2) 

Knowing the inertial coordinates of the parachute center of pressure and the inertial angular velocity, the 

tangential velocity vector (V t ) of the canopy can be determined via Eq. (34). 

! 

# 

V t 

= # 

# 

" 

# 

x cp 

y cp 

z cp 

V t 

= Ω × R cp 

(34.1) 

$ ! 

& 

− 

$ 

# 

θ sinθ cosψ − ψ cosθ sinψ 

& 

& = R cp 

# ψ cosψ & 

& # 

% 

& 

" 

−θ 

& 

# cosθ cosψ + ψ sinθ sinψ 

% 

& 

(34.2) 

The total wind velocity at the canopy is the sum of the blockage-corrected wind velocity and the wind velocity 

due to tangential motion of the canopy. The actual wind velocity vector and magnitude are given in Eq. (35). 

! 

# 

V w 

= # 

# 

"# 

V c 

0 

0 

$ ! 

& # 

& −# 

& # 

%& 

" 

# 

x cp 

y cp 

z cp 

$ ! 

& # 

& = # 

& # 

% 

& # 

" 

V w 

= 

V c 

+ R cp 

( θ sinθ cosψ + ψ cosθ sinψ) 

−R cp 

ψ cosψ 

R cp 

( θ cosθ cosψ − ψ sinθ sinψ) 

( V c 

− x cp ) 2 + y 2 2 

cp 

+ z cp 

$ 

& 

& 

& 

& 

% 

(35.1) 

(35.2) 

Acknowledgments 

This work was directed by the Jet Propulsion Laboratory, California Institute of Technology, under a contract 

with the National Aeronautics and Space Administration. The authors would like to thank Mark Schoenenberger and 

Juan Cruz. Their insight on the data reduction improvements and final results was invaluable. 

References 

1 Tanner, C., Clark, I., Gallon, J., Rivellini, T., and Witkowski, A., “Aerodynamic Characterization of New Parachute 

Configurations for Low-Density Deceleration,” 22 nd AIAA Aerodynamic Decelerator Systems Conference, Daytona Beach, CA, 

March 2013. 

2 Cruz, J. R., Mineck, R., Keller, D., and Bobskill, M., “Wind Tunnel Testing of Various Disk-Gap-Band Parachutes,” 

19 




AIAA Aerodynamic Decelerator Systems Technology Conference and Seminar, AIAA 2003-2129, May 2003. 

3 

Schoenenberger, M., Queen, E. M., and Cruz, J. R., “Parachute Aerodynamics from Video Data,” AIAA Aerodynamic 

Decelerator Systems Technology Conference and Seminar, AIAA 2005-1633, May 2005. 

4 Mitcheltree, R., Burno, R., Slimko, E., Ba_es, C., Konefat, E., and Witkowski, A., “High Altitude Test Program for a 

Mars Subsonic Parachute,” AIAA Aerodynamic Decelerator Systems Technology Conference and Seminar, AIAA 2005-1659, 

May 2005. 

5 Kushner, L.K. and Schairer, E.T., “Planning Image-Based Measurements in Wind Tunnels by Virtual Imaging,” AIAA 

Paper 2011-0930, presented at 49 th AIAA Aerospace Sciences Meeting and Exhibit, Orlando, FL, Jan. 2011. 

6 Abdel-Aziz, Y.I. and Karara, H.M., “Direct Linear Transformation from comparator coordinates into object-space 

coordinates,” Proceedings, Symposium on Close-Range Photogrammetry, American Society of Photogrammetry, Urbana, IL, Jan. 

1971. 

7 McBride, D. D., “Orbiter Drag Chute Stability Test in the NASA Ames 80x120 Foot Wind Tunnel,” Tech. Rep. SAND- 

93-2544, Sandia National Laboratories, Albuquerque, NM, Feb. 1994. 

8 

Cockrell, D., Doherr, K. F., “Preliminary Consideration of Parameter Identification Analysis from Parachute Aerodynamic 

Flight Test Data,” AIAA Paper 1981-1940, AIAA Aerodynamic Decelerator and Balloon Technology Conference, Oct. 1981. 

9 

Ibrahim, S. K., “Experimental Determination of the Apparent Moment of Inertia of Parachutes," Tech. Rep. AFFDL- 

TDR-64-153, Air Force Flight Dynamics Laboratory, Wright-Patterson Air Force Base, Ohio, April 1965. 

20

Aerodynamic Stability and Performance of Next-Generation ...

You also want an ePaper? Increase the reach of your titles

Delete template?

Save as template?